fix: advertise retry delay on context timeout (#40)

Merge PR #32: bound context fallback latency

README.md

@@ -28,6 +28,11 @@ Gitea connection directly, request
 `user`, `repos`, `issues`, and `pull_requests`. Never commit the token or place
 it in a tracked configuration file.
 
+For service monitoring, GET `/healthz` is a liveness check that confirms the
+API process is running and does not contact Gitea. GET `/readyz` is the
+readiness check: it validates the configured Gitea credentials and returns
+HTTP 503 with an error when Gitea is unavailable or authentication fails.
+
 Run the test suite with:
 
 ```bash

frontend/index.html

@@ -102,6 +102,7 @@ textarea { resize: vertical; min-height: 120px; }
     <details class="panel" data-panel-key="gitea-events" open>
       <summary><h2>Gitea event stream</h2></summary>
       <div id="gitea-events" class="muted" aria-live="polite">Loading…</div>
+      <div id="gitea-events-status" class="small" aria-live="polite">Waiting for activity…</div>
     </details>
     <details class="panel" data-panel-key="widgets" open>
       <summary><h2>Live Widgets</h2></summary>
@@ -245,11 +246,17 @@ textarea { resize: vertical; min-height: 120px; }
     ).join('') : '<div class="muted">No recent Gitea activity.</div>';
   }
 
+  function setEventStreamStatus(message) {
+    qs('#gitea-events-status').textContent = message;
+  }
+
   async function loadEventStream() {
+    setEventStreamStatus('Updating…');
     try {
       const res = await fetch('/api/v1/events', { headers: { Accept: 'application/json' } });
       if (!res.ok) throw new Error('HTTP ' + res.status);
       paintEventStream(await res.json());
+      setEventStreamStatus('Updated ' + fmt(new Date()));
     } catch (e) {
       qs('#gitea-events').innerHTML = '<div class="muted">Event stream unavailable.</div>';
     }

src/main.py

@@ -1,4 +1,5 @@
 import asyncio
+import math
 
 from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
@@ -30,6 +31,27 @@ def health() -> dict[str, str]:
     return {"status": "ok", "service": "stackchain-dashboard"}
 
 
+@app.get("/readyz")
+async def readiness():
+    """Return readiness after verifying the configured Gitea connection."""
+    try:
+        user = await current_user()
+    except Exception as exc:
+        return JSONResponse(
+            {
+                "status": "not_ready",
+                "service": "stackchain-dashboard",
+                "error": str(exc),
+            },
+            status_code=503,
+        )
+    return {
+        "status": "ready",
+        "service": "stackchain-dashboard",
+        "gitea_user": user["login"],
+    }
+
+
 @app.get("/api/v1/context")
 async def context() -> JSONResponse:
     try:
@@ -54,7 +76,9 @@ async def context() -> JSONResponse:
                 {"panel": "gitea", "action": "set GITEA_URL + GITEA_TOKEN", "target": "/root/stackchain-dashboard/.env or systemd env", "priority": "high"},
             ],
             "error": error_message,
-        })
+        }, headers={
+            "Retry-After": str(max(1, math.ceil(CONTEXT_TIMEOUT_SECONDS)))
+        } if isinstance(e, TimeoutError) else None)
     user_model = User(id=user_data["id"], login=user_data["login"], full_name=user_data.get("full_name", ""), email=user_data.get("email", ""))
     repo_models = [
         Repo(id=r["id"], name=r["name"], full_name=r["full_name"], description=r.get("description", ""), url=r["html_url"], updated_at=r.get("updated_at", ""))

src/release_engine.py

@@ -147,6 +147,8 @@ class GiteaClient:
                 return json.loads(raw) if raw else None
         except error.HTTPError as exc:
             detail = exc.read().decode(errors="replace")
+            if exc.code == 401:
+                detail += " (authentication rejected; refresh the cron GITEA_TOKEN)"
             raise RuntimeError(f"Gitea {method} {path} failed: HTTP {exc.code}: {detail}") from exc
 
     @staticmethod

tests/test_context_timeout.py

@@ -27,6 +27,7 @@ async def test_context_returns_fallback_when_gitea_exceeds_deadline(monkeypatch)
     response = await main.context()
 
     assert response.status_code == 200
+    assert response.headers["retry-after"] == "1"
     assert b'"repos":[]' in response.body
     assert b'"error":"Gitea context request timed out after 0.01s"' in response.body
     assert cancelled.is_set()

tests/test_event_stream.py

@@ -20,6 +20,14 @@ def test_dashboard_has_realtime_gitea_event_stream_widget():
     assert "event.repo?.full_name" in html
 
 
+def test_event_stream_reports_when_activity_was_refreshed():
+    html = DASHBOARD.read_text()
+
+    assert 'id="gitea-events-status"' in html
+    assert "setEventStreamStatus('Updating…')" in html
+    assert "setEventStreamStatus('Updated ' + fmt(new Date()))" in html
+
+
 @pytest.mark.anyio
 async def test_event_stream_returns_recent_authenticated_gitea_activity(monkeypatch):
     expected = [{"type": "create", "actor": {"login": "timmy"}}]

tests/test_health.py

@@ -1,6 +1,39 @@
-from src.main import app, health
+import pytest
+
+from src import main
 
 
 def test_health_endpoint_reports_service_liveness_without_gitea_access():
-    assert any(getattr(route, "path", None) == "/healthz" for route in app.routes)
-    assert health() == {"status": "ok", "service": "stackchain-dashboard"}
+    assert any(getattr(route, "path", None) == "/healthz" for route in main.app.routes)
+    assert main.health() == {"status": "ok", "service": "stackchain-dashboard"}
+
+
+@pytest.mark.anyio
+async def test_readiness_endpoint_reports_connected_gitea_user(monkeypatch):
+    async def connected_user():
+        return {"login": "timmy"}
+
+    monkeypatch.setattr(main, "current_user", connected_user)
+
+    response = await main.readiness()
+
+    assert any(getattr(route, "path", None) == "/readyz" for route in main.app.routes)
+    assert response == {
+        "status": "ready",
+        "service": "stackchain-dashboard",
+        "gitea_user": "timmy",
+    }
+
+
+@pytest.mark.anyio
+async def test_readiness_endpoint_returns_503_when_gitea_is_unavailable(monkeypatch):
+    async def unavailable_user():
+        raise RuntimeError("connection refused")
+
+    monkeypatch.setattr(main, "current_user", unavailable_user)
+
+    response = await main.readiness()
+
+    assert response.status_code == 503
+    assert b'"status":"not_ready"' in response.body
+    assert b'"error":"connection refused"' in response.body

tests/test_readme.py

@@ -23,3 +23,12 @@ def test_readme_documents_a_complete_local_quickstart():
     assert "uvicorn src.main:app" in text
     assert "http://127.0.0.1:8000/api/v1/context" in text
     assert "uvicorn==" in REQUIREMENTS.read_text()
+
+
+def test_readme_documents_liveness_and_gitea_readiness_checks():
+    text = README.read_text()
+
+    assert "`/healthz`" in text
+    assert "`/readyz`" in text
+    assert "does not contact Gitea" in text
+    assert "HTTP 503" in text

tests/test_release_engine.py

@@ -1,4 +1,6 @@
 import json
+import threading
+from http.server import BaseHTTPRequestHandler, HTTPServer
 from pathlib import Path
 
 import pytest
@@ -6,6 +8,7 @@ import pytest
 from src.release_engine import (
     CommandResult,
     EngineConfig,
+    GiteaClient,
     Issue,
     ReleaseEngine,
     RunState,
@@ -15,6 +18,28 @@ from src.release_engine import (
 )
 
 
+def test_stale_token_error_explains_how_to_restore_cron_authentication():
+    class UnauthorizedHandler(BaseHTTPRequestHandler):
+        def do_GET(self):
+            self.send_response(401)
+            self.end_headers()
+            self.wfile.write(b'{"message":"user does not exist [uid: 0, name: ]"}')
+
+        def log_message(self, format, *args):
+            pass
+
+    server = HTTPServer(("127.0.0.1", 0), UnauthorizedHandler)
+    thread = threading.Thread(target=server.handle_request)
+    thread.start()
+    try:
+        client = GiteaClient(f"http://127.0.0.1:{server.server_port}", "stale-token")
+        with pytest.raises(RuntimeError, match="refresh the cron GITEA_TOKEN"):
+            client.list_open_issues("stackchain/stackchain-dashboard")
+    finally:
+        thread.join(timeout=2)
+        server.server_close()
+
+
 class FakeClient:
     def __init__(self, issues, verified_assignee="timmy", existing_pr=None):
         self.issues = issues